SCIENCE CHINA

Technological Sciences
RESEARCH PAPER

May 2013 Vol.56 No.5: 12571272

doi: 10.1007/s11431-013-5202-9

Unanswered questions in unsaturated soil mechanics

SHENG DaiChao1, 2, ZHANG Sheng1* & YU ZhiWu1
1

School of Civil Engineering, Central South University, Changsha 410075, China;

2
The University of Newcastle, NSW 2308, Australia

Received January 5, 2013; accepted February 22, 2013; published online April 4, 2013

The last two to three decades have seen significant advances in the mechanics of unsaturated soils. It is now widely recognized that the fundamental principles in soil mechanics must cover both saturated and unsaturated soils. Nevertheless, there is
still a great deal of uncertainties in the geotechnical community about how soil mechanics principles well-established for saturated soils can be extended to unsaturated soils. There is even wide skepticism about the necessity of such extension in engineering practice. This paper discusses some common pitfalls related to the fundamental principles that govern the volume
change, shear strength and hydromechanical behaviour of unsaturated soils. It also attempts to address the issue of engineering
relevance of unsaturated soil mechanics.
unsaturated soils, constitutive modeling, volume change, shear strength, engineering applications
Citation:

Sheng D C, Zang S, Yu Z W. Unanswered questions in unsaturated soil mechanics. Sci China Tech Sci, 2013, 56: 12571272, doi:
10.1007/s11431-013- 5202-9

1 Introduction
Research on unsaturated soil behaviour probably dates back
to 1920s [1], but it was not until 1950s that significant research effort has been devoted to the geotechnical aspects of
unsaturated soils [29]. Some early Chinese contributions
towards understanding unsaturated soil behaviour include,
for example, the classification of liquid and gas status in
unsaturated soils by Yu & Chen [10], and the stress-strainstrength behaviour of unsaturated soils by Lu et al. [11],
Chen et al. [12, 13], Shen [14] and Li et al. [15]. Some early
Japanese contributions include the theoretical work of
Karube et al. [16] and Kohgo et al. [17, 18]. The subject of
unsaturated soil mechanics has particularly flourished since
1990s, much attributed to the inspirational work of Alonso
et al. [19] and Fredlund et al. [20]. Indeed, the most significant developments in theoretical soil mechanics during the

*Corresponding author (email: zhsh1230@126.com)

Science China Press and Springer-Verlag Berlin Heidelberg 2013

last three decades or so have probably occurred in the area

of unsaturated soil mechanics. Today the subject is still one
of the most active and prolific research areas in soil mechanics and geotechnical engineering.
Even though the last two to three decades have seen significant advances in unsaturated soil mechanics, there is still
a great deal of uncertainties in the geotechnical community
about how well-established soil mechanics principles for
saturated soils can be extended to unsaturated soils. There is
even wide reservation about the necessity of such extension
in engineering practice. In particular, some basic questions
are often raised on the fundamental principles that govern
the hydromechanical behaviour of unsaturated soils and on
the engineering relevance:
(1) Reconstituted soil versus compacted soil. What are
the main differences in the hydromechanical behaviour of
these soils? What are the implications of different pore size
distributions (PSD), in constitutive modelling of unsaturated
soils? Can a reconstituted soil become collapsible?
(2) Relationship between volume change, yield stress and
tech.scichina.com

www.springerlink.com

1258

Sheng D C, et al.

Sci China Tech Sci

shear strength. Can the constitutive equations for volume

change, yield stress and shear strength be defined separately? Does the loading-collapse yield surface have to recover the apparent tensile strength surface? Do we need the
suction-increase surface to capture possible plastic volume
change when a soil is dried to a historically high suction?
What are the implications of stress state variables in defining volume change and shear strength equations?
(3) Implications of using a Bishop effective stress. Can
we use a Bishop-type effective stress in modelling unsaturated soil behaviour and what are the implications?
(4) Engineering relevance. What is the relevance of the
unsaturated soil mechanics in engineering practice? Is a
design based on the saturated soil mechanics always conservative? Considering the difficulty and uncertainty in
measuring or monitoring in-situ suctions, the applicability
of the unsaturated soil mechanics to engineering practice
has also been questioned.
These questions represent some of the most fundamental
issues in unsaturated soil mechanics. There are currently no
unified answers to these questions. This paper only represents the authors own understanding of these issues. It is
our intention that the paper can serve as a springboard leading to more in-depth discussion and perhaps more insightful
understanding of the fundamental issues of unsaturated soil
mechanics.

May (2013) Vol.56 No.5

and constitutive models for unsaturated soils are almost

exclusively based on experimental data for compacted soils
[21]. The key question is then: Can we extend these principles or models to other types of unsaturated soils? The answer to this question also affects the applicability of unsaturated soil mechanics to engineering practice. After all,
we want to apply our unsaturated soil mechanics to various
unsaturated soils, not just compacted soils. As pointed out
by Gens et al. [22] and Sheng [21], all soils can be unsaturated with water and partial saturation is only the state of
soil, not a new soil.
To answer the question, we need to understand the main
differences between the different types of soil samples. The
most important difference is perhaps the microstructural
differences. Compacted soil samples can be prepared dry of
optimum or wet of optimum. Reconstituted samples can be
air-dried, heat-dried, freeze-dried or osmotically-dried. Different sample preparation methods usually result in different
soil microstructures. For example, soils compacted dry of
optimum tend to have a double-porosity microstructure,
meaning that the pore size distribution curve exhibits two or
more peaks (Figure 1(b)). In these soils, there usually exist

2 Natural, reconstituted versus compacted soils

Soil is a porous material. The pores in a soil can be filled
with different fluids. The term of unsaturated soil refers to
the state where the pores are filled partly with liquid water
and partly with air. A soil can become unsaturated with water in different ways. In field, soils above the ground water
table are naturally unsaturated. Engineered soils such as
those compacted fills used in road and railway embankments are usually unsaturated. In laboratory, three types of
unsaturated soil samples are commonly used: (1) samples
statically or dynamically compacted from dry soil powders
mixed at specified water contents, (2) samples reconstituted
from slurry (at moisture contents in excess of the liquid limit) and then dried to unsaturated states, and (3) undisturbed
samples that are naturally unsaturated. The first type of
samples (compacted soils) is far more common than the
second and third types of samples (reconstituted soils and
natural soils). One reason for this is that it is much easier to
desaturate a compacted sample than reconstituted sample.
The second type sample is commonly used for saturated soil
testing in laboratory, but not so common for unsaturated soil
testing. Undisturbed samples are difficult to obtain and extremely difficult to duplicate. Because laboratory experiments often require a fair amount of repeated tests on the
same sample, the use of undisturbed samples is very limited.
As a consequence, principles of unsaturated soil mechanics

Figure 1 Pore size characteristics of decomposed granitic soil compacted

(after Li & Zhang [23]). (a) SEM of decomposed granitic soil compacted
dry of optimum; (b) pore size distribution of decomposed granitic soil.

Sheng D C, et al.

Sci China Tech Sci

two types of pores: large inter-aggregates pores which are

collapsible upon wetting and small intra-aggregates pore
which are more stable (Figure 1(a)). Some naturally unsaturated soils such as loess and tropical residual soils can
also have a double-porosity microstructure, as shown in
Figure 2(a) for a gneiss residual soil [24]. On the other hand,
soils air-dried from slurry usually exhibit a unimodal pore
size distribution, at least at low stresses (Figure 2(b)).
Unsaturated soils with a bimodal pore size distribution
are usually collapsible at certain stress ranges. Wetting of
such a soil can collapse the inter-aggregates pores and result
in a unimodal pore size distribution when the soil becomes
saturated, as shown in Figure 1(b). In an elastoplastic constitutive model such as the Barcelona Basic Model (BBM)
developed by Alonso et al. [19], a collapsible soil is typically characterised by a loading-collapse (LC) yield surface
where the yield stress increases with increasing suction.
Suction refers to the matric suction, which is the difference
between pore air pressure and pore water pressure. The LC

May (2013) Vol.56 No.5

1259

yield surface in BBM evolves with stress and/or suction

changes. As shown in Figure 3, the compacted soil at point
A is unsaturated (provided that suction is greater than the air
entry suction) and has a bimodal pore size distribution. At
point B the soil is saturated and can have a unimodal pore
size distribution. Experimental data seem to support such an
evolution of pore structure (Figure 1(b)). The wetting path
A B causes the LC yield surface to evolve from LCA to
LCB, as the interaggregate pores collapse and the soil volume decreases. However, drying the unimodal soil at point
B to point A and then compressing it to point C, i.e. stress
path B A C , should regenerate the bimodal pore size
distribution, because the soil at point C is collapsible again
according to BBM [19]. Some experimental data also support such a development of soil collapsibility [26]. In other
words, the pore size distribution can evolve with stress and
hydraulic paths or stress and suction history. Most constitutive models in the literature are based on data for compacted
soils and hence predict an evolution of the LC yield surface
as shown in Figure 3.
The question is: does this kind of yield surface evolution
apply to reconstituted soils as well? A soil air-dried from
slurry is characterised by a yield surface where the yield
stress decreases with increasing suction (Point A in Figure
4). Since such a soil usually has a unimodal pore size dis-

Figure 3 Evolution of pore size distribution and LC yield surface (s:

difference between air pressure and water pressure, p : net mean stress =
difference between total mean stress and air pressure).

Figure 2 Pore size characteristics of (a) natural residual soil and (b)
reconstituted soil. (a) Pore size distribution of a natural residual soil (Futai
& Almeida [24]); (b) pore size distribution of a soil reconstituted from
slurry (Tarantino [25]).

Figure 4 Evolution of the yield surface for a reconstituted soil according

to the SFG model [27].

1260

Sheng D C, et al.

Sci China Tech Sci

tribution, wetting it under constant stress usually does not

cause volume collapse. However, can such a soil become
collapsible if it is compressed to high stresses? According to
the SFG model [27], the yield surface for an air-dried soil
can evolve into a LC yield surface if the soil is compressed
to sufficiently high stresses (Figure 4), which means that
compressing a unimodal reconstituted soil at unsaturated
states can generate a collapsible soil (stress path A C ).
Is there experimental evidence for such an evolution? Unfortunately there is very few data on reconstituted soils in
the literature. Nevertheless, the classic reference by Jennings and Burland [6] seems to support such an evolution.
Figure 5 is a re-plot from Jennings and Burland [6] for an
air-dry silt. It is clear that the air-dry soil can become collapsible if it is compressed to sufficiently high stresses.
This collapsibility is purely due to the degradation of soil
stiffness during wetting. In other words, the volume decrease along stress path BAC is smaller than the volume
decrease along path BD, because the saturated soil (BD) is
more compressible than the unsaturated soil (AC). Therefore, wetting the soil from C to D will cause volume decrease, or the so-called wetting-collapse. However, this collapse is not necessarily related to the microstructural change
of the soil, rather to the overall stiffness change of the soil
at different suction levels. Another set of data reported by
Cunningham et al. [28] for a reconstituted silty clay also
seems to suggest that compressing a soil at sufficiently high

Figure 5

May (2013) Vol.56 No.5

suction can make the soil collapsible (Figure 6). Therefore, the available data in the literature seem to support the
evolution of the LC yield surface as suggested by the SFG
model (Figure 4). However, experimental data on reconstituted soils are generally too few to be conclusive on this
specific characteristic.
In summary, the microstructure and particularly the pore
size distribution of a soil are usually reflected in the yield
surface and volume change behaviour of the constitutive
model for the soil. As recently pointed out by Tarantino
[25], the boundary between compacted and reconstituted
soils is not always clear and the microstructure of a soil can
change with stress and hydraulic paths. A bimodal pore size
distribution can evolve to a unimodal pore size distribution
under appropriate stress and hydraulic paths, and vice versa.
Some existing constitutive models for unsaturated soils can
potentially be applied to different types of unsaturated soils,
at least as a first-order approximation.

3 Relationship between volume change, yield

stress and shear strength
Two classical issues in geotechnical engineering are settlement and stability, respectively related to volume change
and shear strength behaviour of geomaterials. Therefore, volume change and shear strength criteria represent the most

Oedometer curves for air-dry silt soaked at various constant applied pressures (after Jennings & Burland [6]).

Sheng D C, et al.

Sci China Tech Sci

Figure 6 Isotropic compression curves for a reconstituted silty clay at

various suctions ([28]). Sat, saturation; comp, compression; iso, isotropic.

fundamental principles in soil mechanics. For saturated soils,

these criteria are usually described in terms of effective
stress. For unsaturated soils they are also functions of suction and/or degree of saturation. Indeed, two fundamental
issues in unsaturated soil mechanics are the volume change
and shear strength behaviour associated with suction or water content changes. For fine-grained soils like clay and silt
and for soft rocks like claystone and mudstone, increasing
the degree of saturation can cause significant strength loss
and volume change, and these changes of strength and
volume are essentially responsible for the engineering
problems such as rainfall induced landslides, foundation
failures due to wetting-induced strength loss or excessive
deformation, sinkholes caused by water infiltration or seepage, settlement or instability of compacted fill embankments,
etc.
One key feature in unsaturated soil behavior is that the
volume change criterion is closely related to the yield
stress-suction and shear strength-suction relationships [19,
2932]. The so-called LC yield surface which defines the
yield stress-suction relationship is derived from the volume
change equation, as done in the BBM [19] and in the SFG
model [28]. The shear strength-suction relationship can also
be derived from the volume change equation, as shown in
ref. [30]. However, these relationships are often overlooked
in the literature, leading to inconsistent constitutive equations.
The volume change equation usually defines the volume
change caused by a stress or suction change, and is used to
derive the LC surface. The shear strength equation of an
unsaturated soil is expressed through the apparent tensile
strength function and the friction angle of the soil. The apparent tensile strength is usually a function of suction and
this function can be derived from the volume change equation. The apparent tensile strength surface in the stress-suction space also represents the zero shear surface, as the soil

May (2013) Vol.56 No.5

1261

has no shear strength when suction and stress change along

this surface. For saturated soils where the effective stress
principle holds, this surface follows the 135o line in the net
mean stress suction space, and the 90o (vertical) line in the
effective mean stress suction space. For unsaturated soils,
an increment in suction does not usually have the same effect on the soil shear strength as an equal increment in the
mean stress, and hence the zero shear strength will drift
away from 135o or 90o line. In fact, the existing effective
stress definitions for unsaturated soils in the literature are
essentially the definitions of this zero shear strength surface.
The discussion below will show that the zero shear strength
surface is also the initial yield surface of a soil that has a
zero preconsolidation stress and that the LC yield surface
must recover this zero shear strength surface if the preconsolidation stress is set to zero.
The relationship between the shear strength and the LC
yield surface has often been overlooked in some existing
models for unsaturated soils. In the BBM [19] and early
models based on the BBM, the volume change equation has
a singularity when the suction becomes zero. As a consequence, the LC yield surface becomes undefined when the
preconsolidation stress at zero suction is zero, and hence the
apparent tensile strength surface cannot be derived from the
volume change equation. Therefore, a separate apparent
tensile strength surface was introduced in the BBM. However, it should be noted that this is solely due to the singularity in the BBM volume change equation. In later models
based on the Bishop effective stress and in the SFG model,
the volume change equation is continuously defined through
saturated and unsaturated regions. As such, the apparent
tensile strength surface is automatically recovered from the
LC yield surface by setting the preconsolidation stress ( pc0
or pc0 ) to zero at zero suction (Figures 7(a) and 7(b)).
Hence a separate function for the apparent tensile strength
surface is not required. In fact, the original suction-increase
(SI) yield surface in the BBM, which is a horizontal line in
the p s space is used to capture the possible plastic
volume change when a soil is dried to a historically high
suction. The SI yield surface is not needed if the volume
change equation is properly defined, as shown in Figures
7(a) and 7(b). On the other hand, if the LC yield surface
does not recover the apparent tensile strength surface (as
shown in Figures 7(c) and 7(d)) when the preconsolidation
stress ( pc0 or pc0 ) is set to zero, the yield stress and shear
strength functions would become non-unique. The LC yield
surfaces shown in Figures 7(c) and 7(d) are also in conflict
with the definition of yield surface, which are the contours
of the hardening parameter in the stress space. Obviously,
the hardening parameter (e.g., the plastic volumetric strain)
cannot be constant along these LC yield surfaces. The
hardening parameter at points B and C should have the exactly same value as point A in Figures 7(c) and 7(d), but
plastic volumetric strain must occur in order to expand the

1262

Sheng D C, et al.

Sci China Tech Sci

May (2013) Vol.56 No.5

Figure 7 Evolution of yield surface for a reconstituted soil. s, Suction; p , net mean stress; p , effective mean stress; pc0, preconsolidation stress at zero
suction; ABC, stress path under constant net mean stress; CD, stress path under constant suction.

yield surface from B to C. Therefore, the LC yield surface

must recover the apparent tensile strength surface when the
preconsolidation stress at zero suction is set to zero, irrespective of the stress space used.
It is sometimes observed that the shear strength of an
unsaturated soil exhibits a peak value at an intermediate
suction level (Figure 8(a)). Such a peak shear strength has
implications in the volume change equation and in the yield
surface as well. To capture peak shear strength, the equation
that defines the volume change caused by suction changes
should predict a minimum value at the intermediate suction
(Figure 8(b)). As a consequence, the apparent tensile strength should also change (Figure 8(c)) accordingly.
The shear strength of an unsaturated soil is often considered to be sufficiently defined by a single effective stress
[33, 34]). However, this is only true when the friction angle
of the soil does not change with suction. In this case, the
two-stress-variable equation proposed by Fredlund et al.
[35] is equivalent to the single-stress-variable equation
based on an effective stress:

where is the shear strength, c is the effective cohesion

for saturated states, n and n are respectively the effective and net normal stress on the failure plane, and
is the friction angle of the soil.
In eq. (1), the Bishop effective stress parameter () is set
to tan b / tan . It is clear that the variable (s) cannot be
eliminated from eq. (1) if the friction angle of the soil ( )

c n tan s tan b

tan b
c n s
tan

tan c n tan ,

(1)

Figure 8

Implications of peak shear strength at intermediate suctions.

Sheng D C, et al.

Sci China Tech Sci

depends on suction, no matter how the effective stress is

defined. While there are experimental data supporting that
the slope of the critical state line is independent of suction
(e.g., refs. [36, 37]), there is perhaps equal amount of data
supporting the opposite (e.g., refs. [24, 38, 39]). Therefore,
there is no conclusive evidence that the shear strength of an
unsaturated soil can be sufficiently defined by a single effective stress.

4 Implications of using a Bishop effective stress

An important issue about constitutive modelling of unsaturated soils is the choice of the stress space where the constitutive model is built. Early models were usually established in the net stress and suction space (e.g. Alonso et al.
[19]), while the second generation models have mostly used
Bishop effective stress and suction (e.g., Wheeler et al. [40];
Sheng [21]; Tamagnini [41]). The more recent SFG model
(Sheng et al. [27]) has reverted back to the net stress and
suction space.
Net stress and suction are both independent and controllable variables in laboratory tests. Therefore, it is straight
forward to represent laboratory tests using models built in
such a space. The definitions of net stress and suction are
independent of material states and hence the stress space
remains fixed. This is a significant advantage of the independent stress variables. As pointed out by Morgenstern
[42], we normally link equilibrium considerations to deformations through constitutive behaviour and should not introduce constitutive behaviour into the stress state variables.
However, models based on net stress and suction usually
have difficulties in (1) handling the transition between saturated and unsaturated states, (2) handling the shear strength
variation with suction, (3) showing the effects of degree of
saturation and the coupling with hydraulic hysteresis. One
exception is perhaps the more recent SFG model [27],
which seems to have overcome most of the difficulties.
These difficulties are indeed the main motivation for the
shift to the Bishop effective stress in the second generation
models.
The Bishop effective stress is not a controllable variable
in laboratory tests. The effective stress definition usually
involves material state variables such as the degree of saturation or air entry value, and hence the stress space changes
as the material state changes. The constitutive behaviour of
the material is embedded both in the constitutive equation
and the effective stress definition, leading to less explicit
physical meaning of the constitutive law. Nevertheless,
models built in the Bishop effective stress space have many
advantages, for example, the smooth transition between
saturated and unsaturated states, coherent relationship between volume change, shear strength and yield surface, easy
incorporation of the effects of hydraulic hysteresis and the
degree of saturation, a straightforward finite element im-

1263

May (2013) Vol.56 No.5

plementation.
The above-mentioned advantages and shortcomings of
the two categories of models are relatively well known in
the literature. However, there is one specific issue of using
the Bishop effective stress, which is related to the volume
change equation and is less known in the literature, as described below.
A general form of the effective stress can be
p p ( s, S r ) s,

(2)

where p is the net mean stress, p is the effective mean

stress, is the Bishop effective stress parameter and
treated either as a function of suction or as a function of
suction and degree of saturation. Obviously, such a definition of effective stress is very general and covers all the
existing definitions we have used in the literature (perhaps
not Vlahini et al. [43]).
With such an effective stress, the volume change equation for normally consolidated soils is usually assumed to
take the following form:
v N ( s ) ( s ) ln p ,

(3)

where v is the specific volume of the soil, is the compression index, and N is the specific volume when ln p 0 or
the specific volume for a soil slurry (at the liquid limit). If
the effective stress is indeed effective in controlling volume
change, the specific volume (v) should remain constant under constant p , and hence N and should be independent
of suction. However, this is seldom true in reality. In the
literature, N and are usually assumed to be functions of s.
We will show below that a varying N can lead to inconsistency with the yield surface evolution and cannot be
recommended.
In all the models that are based on effective stress, the
apparent tensile strength surface or the zero shear strength
surface is assumed to be the vertical line that goes through
zero effective mean stress, i.e.,
p 0.

(4)

In other words, the soil is assumed to have zero shear

strength as long as the effective mean stress remains zero,
or the soil has no apparent tensile strength in the effective
stress space. This line must also represent the yield surface
for a slurry soil that has never been consolidated. For a saturated soil that has a zero yield stress, i.e. a slurry, the size
of the elastic zone remains zero as long as the effective
mean stress remains zero, irrespective of the pore water
pressure. When the soil becomes unsaturated, the yield surface will continue along the zero shear strength line, if the
effective mean stress remains zero. To keep the effective
mean stress zero, a tensile net mean stress has to be applied
to balance out the suction increase. As a consequence, the
size of the elastic zone remains zero, which also reflects the

1264

Sheng D C, et al.

Sci China Tech Sci

effective stress principle. Furthermore, if the yield surface

does not collapse to the zero shear strength line when
=0 (the dashed LC curve in Figure 9(a)), there would be
pc0
plastic deformation for loading along the yield surface, and
the shear strength as well as the yield stress would become
non-uniquely defined.
Therefore, drying a slurry soil under zero effective mean
stress is a neutral loading process, meaning that the stress
point is always on the yield surface (path AB in Figure
9(a)). The elastic zone does not expand, which is different
from the case of drying a slurry soil under zero net mean
stress (Figure 7(b)). Such a loading process is also logical if
there exists an effective stress. To keep the effective mean
stress zero as the suction is increased, a tensile net mean
stress has to be applied. This tensile net mean stress balances out the effect of the suction increment, which is the essence of the effective stress principle.
As a consequence, drying a slurry soil under zero effective mean stress implies that the soil is always on the normal compression line (NCL). This is only possible: (1) if N
remains constant or decreases with increasing suction (the
dashed line in Figure 9(b)), or (2) if N increases with increasing s (the dotted line in Figure 9(b)).

May (2013) Vol.56 No.5

Case (1) would lead to the constraint that must decrease with increasing suction, which suggests that the collapse strain increases with increasing stress level and the
soil compressibility decreases with increasing suction. Such
behaviour is not always supported by experimental data.
Experimental data by, for example, Toll [38], Sharma [44],
Sivakumar et al. [45] and Toll et al. [46] all indicate that the
soil compressibility can increase with increasing suction
(Figure 10). The data by, for example, Vilar et al. [47], Sun
et al. [48, 49] and Vilar et al. [50] show that the collapse
potential reaches a maximum at intermediate stress levels.
Case (2) suggests that the soil volume increases when drying under constant effective mean stress, which is not supported by any experimental data.
A possible strategy to overcome the above limitation of
the Bishop effective stress approach is to adopt a volume
change equation of the following form:
v N ( S r ) ln p ,

(5)

where N is a constant, and is a function of degree of saturation (Sr).

Because the compression index () is a function of Sr, the
normal compression line under a constant suction would not
be a straight line unless the suction is zero (Figure 11(a)).
Compressing an unsaturated soil under constant suction will
generally lead to an increase in degree of saturation. If the
compression index () is assumed to vary with degree of
saturation according to Figure 11(b), the normal compression line (NCL) for constant suction would take the form of
Figure 11(a), which mimics the oedometer curves for a silty
sand tested by Jennings and Burland [43].
The implication of using a saturation-dependent compression index is that the yield surface will also depend on
Sr. In fact, the LC yield surface can be derived from eq. (5):
(1)

pc pc0 ( Sr ) ,

(6)

where pc is the yield stress, pc0 is the yield stress at

Figure 9 Inconsistency of a varying N in the Bishop effective stress

approach.

Figure 10

Variation of with degree of saturation (after Toll [38]).

Sheng D C, et al.

Sci China Tech Sci

May (2013) Vol.56 No.5

1265

constant suction (CD in Figure 8(c)) in the p Sr space.

Nevertheless, there are a few advantages of this approach:
(1) the constraints on the compression index due to the use
of the Bishop effect stress are avoided; (2) it is a natural
outcome of eq. (5) that the collapsible volume reaches a
maximum value at an intermediate stress level (Figure
11(a)); (3) according to such a model, it is possible to compress an unsaturated soil to full saturation even if the suction is kept constant.
The above approach was initially suggested by Sheng
[27] and a complete model has recently been developed by
Zhou et al. [5154]. The model by Zhou et al. shows very
promising features in capturing unsaturated soil behaviour.
The challenge of using this approach is that simple stress
paths in laboratory tests become very complex and a complete mathematical model is always needed to interpret laboratory tests, in addition to the other implications of the
Bishop effective stress mentioned above.
Another issue related to the Bishop effective stress is that
a closed loop of net stress and suction changes do not necessarily lead to a closed loop of effective stress changes
(Figure 12), because of the material state dependency of the
Bishop effective stress. For example, the effective mean
stress change along path AB is usually different from that
along path CD, because the degree of saturation has
changed from A to C, even though the net mean stress is
kept constant during both paths (Figure 12). Such an unclosed loop means that the model is stress-path dependent,
even if the stress changes are inside the elastic zone. In a
recent discussion by Zhang et al. [55] and Sheng et al. [30],
it was found that models based on net stress were usually
stress-path dependent in the elastic zone, but the discussion
did not realise that models based on effective stress were
also stress-path dependent. A model that exhibits stress-path
dependent elastic behaviour is at variance with classical
elastoplasticity theory and thermodynamics and should
generally be avoided. However, an unsaturated soil model
that exhibits stress-path independent elastic behaviour does
not currently exist in the literature and is a remaining challenge.
Figure 11

Saturation-dependent compression index ( ( Sr ) ) and its

5 Engineering relevance

implications.

saturation (Sr=1), and is the elastic compression index.

Because of the non-unique relationship between Sr and s
due to hydraulic hysteresis, the yield surface can no longer
be plotted in the space of effective mean stress versus suction. Instead, we have to use the degree of saturation as the
additional axis of the stress space, and plot the yield surface
in the space of effective mean stress versus degree of saturation (Figure 11(c)). Because neither the degree of saturation nor the effective stress is a controllable variable in laboratory tests, it will be difficult to represent simple stress
paths like constant net mean stress (ABC in Figure 11(c)) or

Whereas the unsaturated soil mechanics has been one of the

Figure 12 A closed loop in the net stress space and its corresponding
open loop in the effective stress space.

1266

Sheng D C, et al.

Sci China Tech Sci

most active and prolific research areas in soil mechanics

during the last three decades or so, its theoretical development is not without questioning. For example, its engineering relevance has often been questioned. One common perception is that, if we design for fully saturated soils using
the classic soil mechanics principles, we are on the conservative side, making the unsaturated soil mechanics redundant. Another common argument is that, since suction is
difficult to measure and it varies significantly in-situ, it is
not practical to apply the unsaturated soil mechanics principles to engineering practice. We will show, through case
studies, that such perceptions are not correct and unsaturated soil mechanics is indeed very relevant in engineering
practice and can play an essential role in some engineering
problems.
5.1

Stability of unsaturated soil and rock slopes

The loss of shear strength due to the increase in degree of

saturation in unsaturated soil slopes is one of the most
common reasons for landslides, which result in millions of
dollars of socioeconomic cost every year. The Thredbo
landslide, for example, was a disastrous landslide that occurred at the ski resort of Thredbo, New South Wales, Australia in 1997 and resulted in a total of 18 deaths (Figure
13). The New South Wales Government subsequently spent
$40 million in out-of-court settlement with 91 business and
individuals. According to the Supreme Court verdict, the
landslide was mainly caused by the leakage of main water
pipeline which saturated the surrounding soil and in turn
dramatically reduced soil shear strength. Another example
is the Kwun Lung Lau landslide, which occurred in Hong
Kong on 23 July, 1994 and resulted in five fatalities. This
landslide was closely related to the saturation and weakening of the soil mass after a heavy rainfall. Interestingly,
however, rainfall had largely ceased for about 10 hours before the failure. The ground was made of a rather permeable
fill overlying partially weathered volcanic tuff. The forensic
investigation concluded that the landslide was most likely to
have been caused by the ingress of a large volume of water
as a result of leakage from underground services (stormwater pipes and a sewer), bringing about an increase of degree of saturation and a reduction of the shear strength of
the ground [57, 58]).
Another type of instability problems related to unsaturated soils is the slopes excavated in certain rocks such as
claystone, mudstone, marl and siltstone. These types of materials behave more like rock when they are dry and intact,
but become mud or clay when wet or weathered. Excavation
facilitates the formation of fissures and cracks in these materials, and consequently facilitates the infiltration of liquid
water or air moisture into them. Figure 14 shows the typical
degradation process of claystone in water. Exposure to air
moisture (humidity) can also significantly reduce the strength of claystone, as shown in Figure 15. Pineda [60] recent-

May (2013) Vol.56 No.5

ly showed that the strength and stiffness of marls and claystones decreased significantly with increasing number of
humidity change cycles (Figure 16). Chandler [61] and
Berdugo [59] also showed the significant decrease of undrained shear strength of claystones with water content
(Figure 17). This loss of strength and stiffness, combined
with the development of tensile cracks caused by excavation
and cutting, can cause slope instability and landslide problems (Figure 18). For example, landslides occurred along
the F3 freeway between Sydney and Newcastle, due to the
wetting of thin claystone layers and deep tensile cracks.
One key question about landslides of unsaturated soil
slopes is: how can we incorporate the strength reduction due
to saturation increase into the slope stability analysis? Much
progress has been made in this respect. For example, Cai
and Ugai [62] studied the dependency of the safety factor on
the rainfall intensity and duration via a shear strength reduction scheme. Griffiths and Lu [63] adopted an analytical

Figure 13 Thredbo landslide, NSW, Australia, 1997 (image from Geoscience Australia).

Figure 14 Degradation of claystone in water (Pineda et al. [56]). (a) t=0;

(b) t=5 h; (c) t=46 h.

Figure 15 Degradation of claystone in moist air (Pineda et al. [56]). (a) t=0;
(b) t=150 d.

Sheng D C, et al.

Sci China Tech Sci

Figure 18

Figure 16 Shear strength reduction due to cycles of humidity change

(after Pineda et al. [56]).

Shear strength reduction of claystones with water content.

Landslide in claystone.

solution of the unsaturated flow problem to estimate the

evolution of suction in the slope. The suction distribution
was then integrated into elastoplastic finite element program,
by defining Bishops effective stress, to calculate the slope
stability. Huang and Jia [64] and Chen and Liu [65] have
also shown how to incorporate the suction-dependent shear
strength in stability analysis of unsaturated soil slopes.
The most fundamental elements for analyzing unsaturated
soil slopes are (1) the unsaturated shear strength criterion,
(2) the water retention function, (3) the permeability function, and (4) the influx boundary condition at the ground
surface. The unsaturated soil shear strength is usually written as a function of suction or a function of degree of saturation. There are a number of shear strength equations in the
literature and Sheng et al. [21] provided a recent review of
these equations and their performance against experimental
data. To determine the suction and saturation redistribution
inside soil slopes due to rainfall infiltration and evaporation,
the water retention function and unsaturated permeability
function are pivotally important. As mentioned in Section 5,
water retention behaviour (suction and saturation relation) is
affected by soil deformation or the soil initial density. In
cases where deformation is significant, a coupled hydromechanical approach is preferred. The boundary condition at
the ground surface is another important factor and its determination usually involves, unfortunately, a great deal of
uncertainties. Fredlund and Stianson [66] have recently
discussed how to use the existing weather station data to
determine the influx boundary condition, a very promising
step forward.
5.2

Figure 17

1267

May (2013) Vol.56 No.5

Foundations on expansive soils

Attwooll et al. [67] reported a structure damage case owing

to heave of expansive soils. The building is a manufacturing
facility that was completed in Loveland, Colorado 1993 on
clay shale Pierre Formation.
Before construction, the site investigation had indicated
that the subsurface materials are of expansive characteristics.
Therefore, drilled shaft foundations with a structurally supported basement underlain by a crawl space were adopted

1268

Sheng D C, et al.

Sci China Tech Sci

for design.
Significant heave (almost 23 cm) occurred at portions of
the building in the first 3 years. Some early remediation measures were applied, including replacement of the damaged
concrete masonry unit partition walls, cosmetic repairs, removing the lawn and ponds, and installation of deep dewater ing wells. In spite of these efforts, heave continued progressing across the building. In 2005, heave impacted over
three-quarters of the building: maximum heave was about
30.5 cm and over half of the building had raised more than
6 cm.
The damages on the building had been primarily in the
form of shear cracking of concrete masonry unit partition
walls, drywall distress and racked door frames. In addition,
damages also occurred to the drilled shafts caused by lateral
displacement of ground as the bedrock had expanded, which
was the most significant structural distress. At least 10
drilled shafts with 107 cm diameters had sheared at or near
ground line and the relative displacement across the shear
cracks was 5 cm or more (see Figure 19).
The main reason for heave leading to structure damage
can be attributed to the soil expansion due to soil moisture
rising. Moisture profile indicated that soil moisture at heave
areas was much higher than non-heave areas, as shown in
ref. [67]. Boring I-1 was drilled in the basement where the
greatest heave occurred (22.9 cm), and boring I-2 was in the
location where the heave was about 5 cm. Compared with
non-heave area, the soil moisture raised up to 75% at I-1
and up to 25% at I-2 (see Figure 20).
Forensic investigation indicated that the main reasons
leading to the rapid soil moisture raise included: (1) the excavation was backfilled with river sand and gravel, which
allowed infiltrating water to flow promptly to the bottom of
the building; (2) The perched groundwater flowed to the
pond on the west and towards the building to the east via
high permeability backfills (sand and gravel).
5.3 Accumulated settlement in high-speed rail embankment due to seasonal wetting and drying
Cardoso et al. [68] presented an interesting case where seasonal drying and wetting caused undesirable accumulated
settlement for a high speed railway embankment. The em-

Figure 20 Soil moisture profiles for heave and non-heave areas (after
Attwooll et al. [67]).

bankment cross section is shown in Figure 21(a). The subgrade soil is a non-plastic silty sand classified as SM according to the Unified Soil Classification System. The soil
is considered to be an appropriate base material if it is
properly compacted and drained.
The embankment cross section under seasonal drying and
wetting was simulated using the finite element code CODE_
BRIGHT. The BBM (Alonso et al. [19]) was used to represent the behaviour of subgrade soil. The accumulated settlement at point A (Figure 21(a)) is shown in Figure 21(b).
While the seasonal fluctuations after the first 2 years are
minimum, the initial settlement during the first 2 years can
be significant, particularly for high-speed trains which are
more sensitive to vertical displacements than traditional
geotechnical structures.
Jiang and his co-workers [69, 70]) had paid much attention to the calculation of settlement of high-speed rail embankments founded on unsaturated subgrades. They used a
simplified consolidation theory of unsaturated soils and
modified the unsaturated soil permeability according to
soil-water characteristic curves. They also considered various ground improvement methods of the unsaturated subgrades. Their calculations seem to be reasonably accurate
when compared with field measurements.
5.4

Figure 19

Cracked drilled shaft (Attwooll et al. [67]).

May (2013) Vol.56 No.5

Summary of engineering applications

The case studies discussed above only provide some glim-

Sheng D C, et al.

Sci China Tech Sci

May (2013) Vol.56 No.5

1269

ences. Nevertheless, it should be noted that unsaturated soil

mechanics is still at its early stage and there is still a significant gap between the theory and practice. In-situ measurement of soil suction, a key variable in analyzing unsaturated
soil problems, is still a challenging task, in spite of significant advancements in the last two decades [77]. The high
capacity tensiometer developed at Imperial College (e.g.
Ridley et al. [78], see Figure 22) was a welcome to the geotechnical community. Loureno [79] proposed further
modifications to the high capacity tensiometer. These tensiometers are now commercially available and reasonably
affordable. It is likely to see more practical use of these tensiometers in geotechnical engineering.
As pointed out by Xie [80], unsaturated soil mechanics
complements the classical saturated soil mechanics and is
an integral part of modern soil mechanics. It plays an essential role in problems involving expansive soils and collapsible soils, whereas it is generally important for all types of
soils. It is likely we will see further significant developments in both theoretical and experimental unsaturated soil
mechanics in next few decades. These developments will
foster more engineering applications of unsaturated soil
mechanics principles.

Figure 21 Accumulated settlement in high speed rail embankment (after

Cardoso et al. [68]).

pses of the rich literature on engineering applications of

unsaturated soil mechanics (e.g. Alonso et al. [67], Bao [71,
72], Shen [73], Ng et al. [74], Gens [58], Williams [75],
Kong et al. [76]). We should note that a geotechnical design
based on the assumption of full saturation is not always
conservative, particularly when deformation and serviceability are of main concern. For example, embankments and
dams that are made of compacted materials can experience
accumulated deformation under cyclic wetting and drying,
which in turn affects their serviceability [67]. During the
last three decades or so unsaturated soil mechanics has
shown promising engineering relevance and significance in
foundation engineering, slope stability, underground pipelines, retaining structures, dams, pavements and embankments, geo-environmental engineering, and mining engineering. The evergrowing number of international conferences on unsaturated soils is a testimony of its importance.
There is currently at least one major international conference on unsaturated soils every year (the International Conference on Unsaturated Soils-every four years, the AsianPacific Conference-every three or four years, the European
Conference-every four years, and the new established PanAmerican Conference), not to mention specific sessions
dedicated to unsaturated soils in other geotechnical confer-

Figure 22 Imperial College suction probe and tensiometer (after Ridley

et al. [78]).

1270

Sheng D C, et al.

Sci China Tech Sci

6 Concluding remarks
In an attempt to answer those questions posed in the Introduction, the following remarks can be made:
1) It seems possible to use the same theoretical framework to model reconstituted soils and compacted soils. The
pore size distribution evolves with stress and suction paths
and can be modelled by the evolution of the loading collapse yield surface. However, much experimental evidence
is needed before an affirmative conclusion can be drawn. In
particular, we need more experimental data on the behaviour of natural soils or soils reconstituted from slurry.
2) The volume change, yield stress and shear strength
behaviour of an unsaturated soil are co-related to each other
and it is not recommended to define these functions separately. Of all these functions, the volume change equation is
the most fundamental one and it underpins the yield stresssuction and shear strength-suction relations.
3) The loading-collapse yield surface should recover the
apparent tensile strength surface when the preconsolidation
stress at zero suction is set to zero, to avoid non-uniqueness
of the yield surface. The suction-increase yield surface used
to capture the possible plastic volume change associated
with drying is not truly needed, if the loading-collapse yield
surface is properly defined.
4) The shear strength of an unsaturated soil can be defined by a single effective stress, if the friction angle of the
soil does not change with suction. On the other hand, the
volume change behaviour of unsaturated soils usually has to
be expressed in two stress variables.
5) There are implications associated with using the
Bishop effective stress. Because there is only one compression index associated with both stress and suction changes
in the volume change equation, this compression index is
constrained to decrease with increasing suction. Such a constraint is not always supported by experimental data. A possible solution is to adopt a saturation-dependent compression index and to form the constitutive equations in the
stress-saturation space.
6) All the existing elastoplastic models for unsaturated
soils have stress-path dependent elastic behaviour. It is still
a challenging task to solve this theoretical problem.
The mechanics of unsaturated soils has strong engineering relevance. Most geotechnical problems involve some
variations of suction, water content or degree of saturation.
These variations can cause significant volume change and
strength variation, leading to undesirable deformation and
stability problems. It is likely that we will see further developments and more engineering applications of unsaturated soil mechanics in the next few decades.
This work was supported by the National Natural Science Foundation of
China (Grant No. 51208519).

1
2

3
4
5
6
7
8
9
10

11

12

13
14
15

16

17
18
19
20
21
22
23
24

25

26
27

May (2013) Vol.56 No.5

Gardner W, Widtsoe JA. The movement of soil moisture. Soil Sci,

1921, 11(3): 215232
Aitchison G D, Donald I B. Effective stresses in unsaturated soils. In:
Proc. 2nd Australian-New Zealand Conf. Soil Mechanics, Australian,
1956, 192199
Bishop A W. The principle of effective stress. Teknisk Ukeblad,
1959, 106(39): 859863
Bishop A W, Blight G E. Some aspects of effective stress in saturated
and partly saturated soils. Gotechnique, 1963, 13: 177197
Coleman J D. Stress strain relations for partly saturated soil. Gotechnique, 1962, 12: 348350
Jennings J E B, Burland J B. Limitations to the use of effective
stresses in partly saturated soils. Gotechnique, 1962, 12: 125144
Matyas E L, Radhakrishna H S. Volume change characteristics of
partially saturated soils. Gotechnique, 1968, 18: 432448
Fredlund D G, Morgenstern N R. Constitutive relations for volume
change in unsaturated soils. Can Geotech J, 1976, 13(3): 261276
Fredlund D G, Morgenstern N R. Stress state variables for unsaturated soils. J Geotech Eng Div, ASCE, 1977, 103(GT5): 447466
Yu P J, Chen Y Z. The pore air-water configurations and their effects
on the mechanical properties of partially saturated soils (in Chinese).
J Hydraul Eng, 1965, 1: 1623
Lu Z J, Zhang H M, Chen J H et al. Shear strength and swelling
pressure of unsaturated soil (in Chinese). Chin J Geotech Eng, 1992,
14(3): 18
Chen Z H, Xie D Y, Wang Y S. Experimental studies of laws of fluid
motion, suction and pore pressures in unsaturated soil (in Chinese).
Chin J Geotech Eng, 1993, 15(3): 920
Chen Z H, Xie D Y, Wang Y S. Effective stress in unsaturated soil
(in Chinese). Chin J Geotech Eng, 1994, 16(3): 6269
Shen Z J. Generalized suction and unified deformation theory for unsaturated soils (in Chinese). Chine J Geotech Eng 1996, 18(2): 19
Li X K, Fan Y Q. Finite element analysis of deformation and seepage
process in unsaturated soils (in Chinese). Chin J Geotech Eng, 1998,
20(4): 2024
Karube D, Kato S. Yield functions of unsaturated soil. Proceedings of
the 12th International Conference on Soil Mechanics and Foundation
Engineering, vol 1. Leiden: Balkema, 1989. 615618
Kohgo Y, Nakano M, Miyazaki T. Theoretical aspects of constitutive
modeling of unsaturated soils. Soils Found, 1993, 33(4): 4963
Kohgo Y, Nakano M, Miyazaki T. Verification of generalized elastoplastic model for unsaturated soils. Soils Found, 1993, 33(4): 6473
Alonso E E, Gens A, Josa A. A constitutive model for partially saturated soils. Gotechnique, 1990, 40: 405430
Fredlund DG, Rahardjo H. Soil Mechanics for Unsaturated Soils.
New York: John Wiley, 1993
Sheng D, Zhou A N, Fredlund D G. Shear strength criteria for unsaturated soils. Geotech Geol Eng, 2011, 29: 145159
Gens A, Snchez M, Sheng D. On constitutive modelling of unsaturated soils, Acta Geotech, 2006, 1: 137147
Li X, Zhang L M. Characterization of dual-structure pore size distribution of soil. Can Geotech J, 2009, 46: 129141
Futai M M, de Almeida M S S. An experimental investigation of the
mechanical behaviour of an unsaturated gneiss residual soil. Gotechnique, 2005, 55(3): 201214
Tarantino A. Unsaturated soils: compacted with reconstituted states.
In: Alonso E E, Gens A, eds. Unsaturated Soils, vol 1. London: CRC
Press, 2010. 113136
Sharma R. Mechanical behaviour of unsaturated highly expansive
soil. PhD Thesis. Oxford: University of Oxford, 1998
Sheng D, Fredlund D G, Gens A. A new modelling approach for un-

Sheng D C, et al.

28

29

30
31

32

33

34

35
36
37
38
39

40

41
42

43

44

45

46
47

48

Sci China Tech Sci

saturated soils using independent stress variables. Can Geotech J,

2008, 45: 511534
Cunningham M R, Ridley A M, Dineen K, et al. The mechanical behaviour of a reconstituted unsaturated silty clay. Gotechnique, 2003,
53: 183194
Wheeler S J, Karube D. Constitutive modelling. In: Alonso E E,
Delage P, eds. Unsaturated Soils, vol 3. Rotterdam: Balkema; 1996.
13231356
Sheng D. Review of fundamental principles in modelling unsaturated
soil behaviour. Comput Geotech, 2011, 38: 757776
Sheng D, Sloan S W, Gens A. A constitutive model for unsaturated
soils: thermomechanical and computational aspects. Comput Mech,
2004, 33: 453465
Sheng D C, Yang C. Discussion of fundamental principles in unsaturated soil mechanics (in Chinese). Chin J Geotech Engi, 2012, 34(3):
438456
Wheeler S J. Constitutive modelling of unsaturated soils. Keynote
Lecture presented at 4th International Conference on Unsaturated
Soil (unpublished PDF presentation), Arizona, USA, 2006.
Alonso E E, Pereira J M, Vaunat J, et al. A microstructurally based
effective stress for unsaturated soils. Gotechnique, 2010, 60:
913925.
Fredlund D G, Morgenstern N R, Widger A. Shear strength of unsaturated soils. Can Geotech J, 1978, 15: 313321
Ng C W W, Chiu A C F. Behaviour of loosely compacted unsaturated
volcanic soil. J Geotech Geoenviron Eng, 2001, 127: 10271036
Thu T M, Rahardjo H, Leong E C. Critical state behaviour of a compacted silt specimen. Soils Found, 2007, 47: 749755
Toll D G. A framework for unsaturated soil behaviour. Gotechnique,
1990, 40: 3144
Merchn V, Vaunat J, Romero E, et al. Experimental study of the influence of suction on the residual friction angle of clays. In: Toll D G,
Augarde C E, Gallipoli D, et al. eds, Unsaturated Soils: Advances in
Geo-Engineering. London: CRC Press, 2008. 423428
Wheeler S J, Sharma R S, Buisson M S R. Coupling of hydraulic
hysteresis and stress-strain behaviour in unsaturated soils. Gotechnique, 2003, 53: 4154
Tamagnini R. An extended cam-clay model for unsaturated soils with
hydraulic hysteresis. Gotechnique, 2004, 54(3), 223228
Morgenstern N R. Properties of compacted soils. Proceedings of the
6th Pan-American Conference on Soil Mechanics and Foundation
Engineering. Contribution to the Panel Discussion Session IV, Lima,
Peru, vol 3, 1979. 349354
Vlahini I, Jennings H M, Andrade J E, et al. A novel and general
form of effective stress in a partially saturated porous material: the
influence of microstructure. Mech Mater, 2011, 43: 2535
Suriol J, Gens A, Alonso E E. Behavior of compacted soils in suction-controlled oedometer, Proc. 2nd Int. Conf. Unsaturated Soils.
Beijing: International Academic Publishers, 1998. 438443
Sivakumar V, Wheeler S J. Influence of compaction procedure on the
mechanical behaviour of an unsaturated compacted clay. Part 1: Wetting and isotropic compression. Gotechnique, 2000, 50: 359368
Toll D G, Ong B H. Critical state parameters for an unsaturated residual sandy clay. Gotechnique, 2003, 53: 93103
Vilar O M, Davies G I. Collapse behavior analysis of a clayey sand
using different testing procedures. In: Juca J F T, de Campos T M P,
Marinho F A M, eds, Unsaturated Soils, vol 2. Leiden: Balkema,
2002. 571 576
Sun D A, Matsuoka H, Xu Y F. Collapse of compacted clay in suction-controlled triaxial tests. Geotech Test J ASTM, 2004, 27(4):
362370

49
50

51

52

53

54

55

56

57
58
59

60
61
62
63

64

65

66

67

68

69

May (2013) Vol.56 No.5

1271

Sun D A, Sheng D, Xu X F. Collapse behaviour of unsaturated compacted soil. Can Geotech J, 2007, 44: 673686
Vilar O M, Rodrigues R A. Collapse behaviour of soil in a Brazilian
region affected by a rising water table. Can Geotech J, 2011, 48:
226233
Zhou A N, Sheng D. Yield stress, volume change and shear strength
behaviour of unsaturated soils: validation of the SFG model. Can Geotech J, 2009, 46: 10341045
Zhou A N, Sheng D, Carter JP. Modelling the effect of initial density
on soil water characteristic curves, Gotechnique, 2012, 62(8): 669
680
Zhou A N, Sheng D, Sloan S W, et al. Interpretation of unsaturated
soil behaviour in the stress-saturation space, I: Volume change and
water retention behaviour. Comput Geotech, 2012, 43: 178187
Zhou A N, Sheng D, Sloan S W, et al. Interpretation of unsaturated
soil behaviour in the stress-saturation space, II: Constitutive relationships and validations. Comput Geotech, 2012, 43: 111123
Zhang X, Lytton R L. Discussion on 'A new modelling approach for
unsaturated soils using independent stress variables'. Can Geotech J,
2008, 45: 17841787
Pineda J A, De Gracia M, Romero E. Degradation of partially saturated argillaceous rocks: Influence on the stability of geotechnical
structures. In: Buzzi O, Fityus S G, Sheng D, eds. Unsaturated Soils
Experimental Studies in Unsaturated Soils and Expansive Soils.
London: CRC Press, 2010. 449454
Wong H N, Ho K K S. The 23 July 1994 landslide at Kwun Lung
Lau, Hong Kong. Can Geotech J, 1997, 34(6): 825840
Gens A. Soil-environmental interactions in geotechnical engineering.
Gotechnique, 2010, 60: 374
Berdugo I R. Tunneling in sulphate-bearing rocksexpansive phenomena. PhD Thesis. Barcelona: Universitat Politcnica de Catalunya, 2007
Pineda J A. Swelling and degradation of argillaceous rocks. PhD
Thesis. Barcelona: Universitat Politcnica de Catalunya, 2011
Chandler R J. Lias clay: weather processes and their effect on shear
strength. Gotechnique, 1972, 22: 403431
Cai F, Ugai K. Numerical analysis of rainfall effects on slope stability. Int J Geomech, 2004, 4(2): 6978
Griffiths D V, Lu N. Unsaturated slope stability analysis with steady
infiltration and evaporation using elasto-plastic finite elements. Int J
Numer Anal Meth Geomech, 2005, 29: 249267
Huang M S, Jia C Q. Strength reduction FEM in stability analysis of
soil slopes subjected to transient unsaturated seepage. Comput Geotech, 2009, 36: 93101
Chen X, Liu C J. Staged development of finite element methods for
stability of unsaturated soil slopes (in Chinese). Chin J Geotech Eng,
2011, 33(S1): 381384
Fredlund D G, Stianson J. Utilization of weather station data for the
assessment of ground surface moisture flux. In: Jotisankasa A,
Sawangsuriya A, Soralump S, et al., eds. Unsaturated Soils: Theory
and Practice, vol 1. Pattaya: Kasetsart University, 2011. 118
Alonso E E, Olivella S. Unsaturated soil mechanics applied to geotechnical problems. In: Miller G A, Zapata C E, Houston S L, et al.
eds. Unsaturated soils ASCE GSP 147, 2006, 1: 135
Cardoso R, Fernandes V, Ferreira T M. Settlement prediction of high
speed railway embankment considering the accumulation of wetting
and drying cycles. In: Mancuso C, Jommi C, D'Onza F, eds. Unsaturated Soils: Research and Applications, vol 2. Heidelberg: Springer, 2012. 291298
Li A H, Yao Y C, Jiang G. Testing swtudy on design parameter of
settlement for unsaturated soil foundation (in Chinese). J Railway

1272

70

71
72

73
74

75

Sheng D C, et al.

Sci China Tech Sci

Eng Soc, 2012, 160: 610

Wu L J, Jiang G L, Li A. Calculation analysis of unsaturated subsoil
subsidence of high speed railway using PLAXIS software (in Chinese). Railway Eng, 2011, 6: 8688.
Bao C G. Behavior of unsaturated soil and stability of expansive soil
slope (in Chinese). Chine J Geotech Eng, 2004, 26: 115
Bao C G, Zhan L T. Relationship between unsaturated soil behavior
and engineering problems (in Chinese). Chin J Geotech Eng, 2006,
28: 129136
Shen Z J. Exploitation of practical use of unsaturated soil mechanics
(in Chinese). Chin J Geotech Eng, 2006, 28: 256259
Ng C W W, Xu J, Zhan L T. The mega south-to-north water transfer
project in ChinaUnsaturated soil mechanics aspects. In: Buzzi O,
Fityus S G, Sheng D, eds. Unsaturated SoilsTheoretical and Numerical Advances in Unsaturated Soil Mechanics. London: CRC
Press, 2009. 545564
Williams D J. Some mining applications of unsaturated soil mechan-

76
77

78

79
80

May (2013) Vol.56 No.5

ics. In: Jotisankasa A, Sawangsuriy A, Soralump S, et al, eds. Unsaturated Soils: Theory and Practice. Pattaya: Kasetsart University,
2011. 149168
Kong L W, Chen Z H. Advancement in the techniques for special
soils and slopes (in Chinese). China Civil Eng J, 2012, 45: 141161
Delage P, Romero E, Tarantino A. Recent developments in the techniques of controlling and measuring suction in unsaturated soils. In:
Toll D G, Augarde C E, Gallipoli D, et al. eds. Unsaturated Soils:
Advances in Geo-Engineering. London: CRC Press, 2008. 3352
Ridley A M, Dineen K, Burland J B, et al. Soil matrix suction: some
examples of its measurement and application in geotechnical engineering. Gotechnqiue, 2003, 53(2): 241254
Loureno S D N. Suction measurements and water retention in unsaturated soils, PhD Thesis. Durham: Durham University, 2008
Xie D Y. Soil mechanics in 21st century (in Chinese). Chin J Geotech
Eng, 1997, 9(4): 111114